Which DOP value may be most relevant to a particular application is dependent on the

mission and associated accuracy requirements of that mission. (K)HDOP may be most

important for land and open ocean navigation where horizontal position location and

rendezvous are primary mission requirements. (K)XDOP and (K)ADOP may be most

important for air navigation where aircraft spacing is a primary safety consideration. 

(K)PDOP may be most important for aircraft weapons delivery, and (K)TDOP is

obviously most important for time transfer applications.  Note that the DOP values

discussed here are instantaneous estimates of the geometric contribution to error for a

particular location and time.  System accuracy requirements often require estimates of

long term error distributions.

For long term error estimates, the relationship between range error and position

solution error should be determined by computer simulation.  The standard deviation of

the long term position error distribution can be determined by using the standard

deviation of GDOP and the standard deviation of UERE, but the relationship does not

hold true for other probability levels, because the tails of the GDOP and position

solution distributions are not Gaussian.  The most effective method for determining

long term error distributions, for a particular constellation state or set of states, is by

conducting a computer simulation.

Computer simulations can be performed to determine global, regional, or single

location distributions, but they are often complex and time consuming.  If a Monte Carlo

simulation is performed assuming a one metre standard deviation for UERE, the

resultant normalized position error (NPE) distribution can be scaled by any UERE of

interest, and examined at any probability level of interest.  The simulation can iterate

user locations around the globe and satellite orbital locations over time (24 hours) while

simulating GPS receiver calculations to determine the NPE distribution.  While NPE is

analogous to GDOP in that it is a measure of the geometric characteristics of error,

GDOP is an instantaneous measure and NPE is a statis tical measure.  The 95% PPS

and SPS accuracy values given in paragraph 5.1 of "Technical Characteristics of the

Navstar GPS", were determined by an NPE simulation using an optimized 21 satellite

constellation as a surrogate for the average or typical state of the GPS constellation.

It should be emphasized that it may be perfectly valid to translate user accuracy

requirements between different dimensions and probability levels assuming a spherical

error distribution and Gaussian error characteristics, if that is appropriate for the

particular mission or application.  The fact that GPS accuracy performance is

nonspherical and non Gaussian does not impose a similar condition on user

requirements.

3.2  RECEIVER POSITION ACCURACY

As described in paragraph 3.1.2.1, the UEE is independent of the satellite and

Control Segment errors, URE and receiver position accuracy are not.  Therefore,

receiver position accuracy must be specified for conditions of DOP and URE, in

order to isolate the receiver contribution to position accuracy (e.g., UEE, filtering

algorithms, and coordinate trans formations).  Dynamic positioning accuracy

requirements must take into account the effect of  vehicle motion on the filter

accuracy as well.  Laboratory testing must control DOP and URE.  Field testing

must record DOP and URE.  In general, testing is best performed when the system

3 6